Talked to @mazzulab about issue with time-varying weights in LGSSM. Found that the sample code needs to pass through to lgssm_joint_sample in the inference code since the base SSM object doesn't support time-varying parameters in the sample code. Also dug into issues with fitting LGSSM and made some changes to his notebook to address those problems.
Talked to @mazzulab about issue with time-varying weights in LGSSM. Found that the sample code needs to pass through to
lgssm_joint_sample
in the inference code since the base SSM object doesn't support time-varying parameters in the sample code. Also dug into issues with fitting LGSSM and made some changes to his notebook to address those problems.